Applications of Explainable Artificial Intelligence in the Healthcare Sector

<h1 style="margin-bottom:0cm;mso-hyphenate:none">Applications&nbsp;of&nbsp;Explainable&nbsp;Artificial&nbsp;Intelligence&nbsp;in&nbsp;the&nbsp;HealthcareSector<o:p></o:p></h1><h2>Gülhan Toğa¹<o:p></o:p></h2><h3>¹Erciyes University, Engineering Faculty, Industrial EngineeringDepartment, Kayseri, Türkiye<o:p></o:p></h3><h3>[email protected]<o:p></o:p></h3><p class="MsoNormal" align="center" style="text-align:center"><o:p>&nbsp;</o:p> <p class="MsoNormal" style="margin-bottom:0cm"><b><i>Abstract: </i></b>The emergence of big data has significantlyexpanded the information accessible from this data. <a name="_Hlk180075122">Theextensive application of machine learning, particularly deep learning, hasenabled the derivation of meaningful insights from big data sets. </a>Nonetheless,end users encounter several deficiencies in the developed models, includingtransparency, interpretability, and reliability. In domains wheredecision-making processes are essential, such as banking, healthcare, andautonomous systems, these shortcomings pose significant problems."Explainable Artificial Intelligence (XAI)," which is created toovercome these limitations, offers a systematic way to help people comprehendthe model and its judgments as well as how to evaluate the models' consistency.This study provides a most recent literature review on the use and benefits ofXAI in the healthcare. An example of application in healthcare is also providedin the study, which predicts the results of interventional operations onpatients with breast cancer. The survival and mortality outcomes of breastcancer patients after surgery are modeled using with the Random ForestClassifier and Gradient Boosting Classifier, and these models are explainedusing one of the XAI techniques, Local Interpretable Model-agnosticExplanations (LIME). LIME is a technique used to understand the predictions ofmachine learning models. It provides interpretable and locally faithfulexplanations by approximating the model locally with a simpler, interpretablemodel. According to the findings, applying XAI aids in gaining a deepercomprehension of the motivation behind the model's decision-making. <o:p></o:p> <p class="Abstracttext" style="margin-bottom:0cm"><o:p>&nbsp;</o:p> <p class="MsoBodyText2" style="margin-bottom:0cm;line-height:normal;mso-outline-level:1"><b><i>Keywords</i></b><b>:</b><i> </i>ExplainableArtificial Intelligence, Gradient Boosting Classifier, Random ForestClassifier, LIME<o:p></o:p> <p class="MsoNormal"><o:p>&nbsp;</o:p>